Forecast Lead Time Research Articles

Effects of Intraseasonal Oscillation on Timing and Subseasonal Predictability of Mei-yu Onset over the Yangtze River Basin

Abstract The time of rainy season onset is crucial information for policymakers, especially in densely populated regions such as the Yangtze River basin (YRB) in China. In this study, we proposed a new grid-based index to objectively detect mei-yu onset timing using reanalysis data and model predictions, and then we identified the key processes via which intraseasonal oscillation (ISO) affects the YRB mei-yu onset and its subseasonal predictability based on scale-decomposed moisture analysis. Climatologically, propagation of an ISO anticyclonic anomaly toward East China supports the moisture convergence required for rainy season onset over the YRB via interaction with the seasonal-mean moisture component. In the years of early mei-yu onset, the ISO was enhanced earlier in May and favored the moisture convergence anomaly in late May–early June, when the mei-yu started. In contrast, the enhanced ISO and associated moistening processes were observed later in June–early July in the years with delayed onset. The European Centre for Medium-Range Weather Forecasts and National Centers for Environmental Prediction models show skillful prediction of mei-yu onset at forecast lead times of 5–6 pentads, whereas the China Meteorological Administration model has limited skill of 3 pentads. The differences in model prediction skill are related to the accuracy of predicted moisture convergence anomalies induced by the ISO. The prediction bias in mei-yu onset timing (early or delayed) is also connected to bias in the occurrence timing of enhanced intraseasonal perturbations, suggesting the vital role of ISO in YRB mei-yu onset on the subseasonal time scale.

Ensemble Forecasts of Extreme Flood Events with Weather Forecasts, Land Surface Modeling and Deep Learning

Integrating numerical weather forecasts that provide ensemble precipitation forecasts, land surface hydrological modeling that resolves surface and subsurface hydrological processes, and artificial intelligence techniques that correct the forecast bias, known as the “meteo-hydro-AI” approach, has emerged as a popular flood forecast method. However, its performance during extreme flood events across different interval basins has received less attention. Here, we evaluated the meteo-hydro-AI approach for forecasting extreme flood events from headwater to downstream sub-basins in the Luo River basin during 2010–2017, with forecast lead times up to 7 days. The proposed meteo-hydro approach based on ECMWF weather forecasts and the Conjunctive Surface-Subsurface Process version 2 land surface model with a spatial resolution of 1 km captured the flood hydrographs quite well. Compared with the ensemble streamflow prediction (ESP) approach based on initial conditions, the meteo-hydro approach increased the Nash-Sutcliffe efficiency of streamflow forecasts at the three outlet stations by 0.27–0.82, decreased the root-mean-squared-error by 22–49%, and performed better in reliability and discrimination. The meteo-hydro-AI approach showed marginal improvement, which suggested further evaluations with larger samples of extreme flood events should be carried out. This study demonstrated the potential of the integrated meteo-hydro-AI approach for ensemble forecasting of extreme flood events.

Exploring the use of 3D radar measurements in predicting the evolution of single-core convective cells

Object-based radar rainfall nowcasting is a widely used technique for convective storm prediction. Currently, most existing object-based nowcasting methods primarily focus on predicting cell movements, neglecting the temporal evolution of cell properties such as size, shape, and intensity. Incorporating this evolution is critical for improving predictability in convective storms. While previous studies have used three-dimensional (3D) radar observations to capture vertical changes during convective cell formation, these efforts often analyse or reconstruct specific convective events. Integrating 3D radar information into operational object-based radar rainfall nowcasting remains an open challenge. This research addresses this challenge using deep learning (DL) techniques. More specifically, a DL-based prediction model is developed, which uses 2D and 3D cells' properties retrieved from 3D radar reflectivity data at the current time and across the past 15 min to predict the evolution of these properties over the next 15 min. This model could eventually be integrated into existing object-based nowcasting models. A total of 4708 cell lifecycles, extracted from high-resolution (5-min, 1-km, 24 levels at 0.5 km intervals) 3D radar data across the UK, are used to train the model, and a total of 1177 lifecycles are used for testing. The proposed model is shown to predict the evolution of single-core convective cells effectively, including changes in 2D projected geometry and mean 2D and 3D reflectivity. In particular, by incorporating information on the vertical evolution of convective cores, the prediction errors of mean reflectivity (in both 2D and 3D) can be reduced by approximately 50% at 15-min forecast lead time, as compared to a persistence forecast. Keywords: radar, tracking, convective cell, nowcasting, 3D, deep learning, lstm.

Seasonal Characteristics of Forecasting Uncertainties in Surface PM2.5 Concentration Associated with Forecast Lead Time over the Beijing-Tianjin-Hebei Region

Seasonal Characteristics of Forecasting Uncertainties in Surface PM2.5 Concentration Associated with Forecast Lead Time over the Beijing-Tianjin-Hebei Region

Assessment and Ensemble-Based Analysis of the Landfalling Typhoon Muifa (2022)

By considering the uncertainties in the initial field, model physical processes, and lateral boundary conditions, the Shanghai Weather And Risk Model System-Ensemble Prediction System (SWARMS-EN) is constructed. According to the prediction results of typhoon Muifa (2022), the daily track error of SWARMS-EN within 5 days is 70.6 km, 142.2 km, 129.1 km, 174.5 km, and 203.5 km, respectively. When compared with the Typhoon Ensemble Data Assimilation and Prediction System (TEDAPS) and the Global Ensemble Forecast System (GEFS) of the National Centers for Environmental Prediction (NCEP) in homogeneous conditions, SWARMS-EN performs better than TEDAPS within 72 h and better than GEFS beyond 72 h in track forecasting. This indicates an improvement in forecasting accuracy. The ensemble spread within two days is less than the root mean square error (RMSE), according to an analysis of the relationship between ensemble RMSE and spread, which shows that SWARMS-EN has no apparent systematic bias overall. The system has improved the ensemble RMSE and spread, indicating that it can better represent the uncertainty of the forecast and produce more reliable forecasts. Additionally, SWARMS-EN provides the landfall forecast five days in advance. The ensemble-based analysis suggests that the large-scale circulation is the primary factor contributing to the forecast differences among members, and the strong steering flow provides an indication of the landfalling forecast. The analysis of the ensemble characteristics of the initial field indicates that the initial perturbation between the wind field and the temperature field in the dynamically unstable region (such as near a tropical cyclone) exhibits flow dependence, and the small perturbation shows continuity throughout the entire troposphere. The distribution of ensemble spread and disturbance energy exhibited a reasonable growth stage as the forecast lead time increased. Disturbance internal energy dominated the lower troposphere, while the upper troposphere was mainly characterized by disturbance kinetic energy. Disturbance kinetic energy played a leading role in the evolution process. This conclusion further confirms the importance of paying attention to the initial small perturbations near TC in order to optimize the initial perturbation.

A cost-benefit analysis of flood early warning system: Evidence from lower Brahmaputra River Basin, Bangladesh

Bangladesh is a severely flood-affected country that ranks fifth globally. Although it is well established that regional and community-based flood early warning systems (FEWS) may minimize the effects of floods, cost-benefit analyses of FEWS are still limited in Bangladesh. This is the first study to assess the cost-benefit of the FEWS in lower Brahmaputra River basin in Bangladesh which included 1000 household surveys, 32 focus group discussion, and key informant interviews. The findings indicated that during the floods, families considered the FEWS to be useful and trustworthy, enabling them to preserve household assets, agricultural, and livestock expenditures worth of BDT 267713 (USD 2525.59) per household. Based on the different scenarios, the benefit-cost ratio (BCR) ranged from 79.87 to 213. About 92% of the respondents expressed willingness to pay an annual charge of BDT 100.63 (USD 0.94) for five years if the community disaster committees were to handle the current FEWS. This might generate BDT 100637 (USD 949.40) annually, to pay for the system's maintenance and operations. As communities' advance confidence in the FEWS system and lead times are enhanced, FEWS increasingly changes their behaviors over time, leading to enhanced social capital and a broader range of early interventions that lessen preventable loss and damage. By extending the forecast lead time by two days, the present savings could increase by 21.7 times. To increase funding, government and non-government organizations may make decisions based on the findings of the cost-benefit analysis. This study also proposed a FEWS applicable in the local level.

A comparison of two statistical postprocessing methods for heavy‐precipitation forecasts over India during the summer monsoon

AbstractAccurate ensemble forecasts of heavy precipitation in India are vital for many applications and essential for early warning of damaging flood events, especially during the monsoon season. In this study we investigate to what extent Quantile Mapping (QM) and Ensemble Model Output Statistics (EMOS) statistical postprocessing reduce errors in precipitation ensemble forecasts over India, in particular for heavy precipitation. Both methods are applied to day‐1 forecasts at 12‐km resolution from the 23‐member National Centre for Medium Range Weather Forecasting (NCMRWF) global ensemble prediction system (NEPS‐G). By construction, QM leads to distributions close to the observed ones, while EMOS optimizes the ensemble spread, and it is not a priori clear which is better suited for practical applications. The methods are therefore compared with respect to several key aspects of the forecasts: local distributions, ensemble spread, and skill for forecasting precipitation amounts and the exceedance of heavy precipitation thresholds. The evaluation includes rank histograms, Continuous Ranked Probability Skill Scores (CRPSS), Brier Skill Scores (BSS), reliability diagrams, and receiver operating characteristic. EMOS performs best not only with respect to correcting under‐ or overdispersive ensembles, but also in terms of forecast skill for precipitation amounts and heavy precipitation events, with positive CRPSS and BSS in most regions (both up to about 0.4 in some areas), while QM in many regions performs worse than the raw forecast. QM performs best with respect to the overall local precipitation distributions. Which aspects of the forecasts are most relevant depends to some extent on how the forecasts are used. If the main criteria are the correction of under‐ or overdispersion, forecast reliability, match between the forecasted distribution for individual days and observations (CRPSS), and the skill in forecasting heavy‐precipitation events (BSS), then EMOS is the better choice for postprocessing NEPS‐G forecasts for short lead times.

Comparison of Clustering Approaches in a Multimodel Ensemble for U.S. East Coast Cold Season Extratropical Cyclones

Abstract Several clustering approaches are evaluated for 1–9-day forecasts using a multimodel ensemble that includes the GEFS, ECMWF, and Canadian ensembles. Six clustering algorithms and three clustering spaces are evaluated using mean sea level pressure (MSLP) and 12-h accumulated precipitation (APCP) for cool-season extratropical cyclones across the Northeast United States. Using the MSLP cluster membership to obtain the APCP clusters is also evaluated, along with applying clustering determined at one lead time to cluster forecasts at a different lead time. Five scenarios from each clustering algorithm are evaluated using displacement and intensity/amount errors from the scenario nearest to the MSLP and 12-h APCP analyses in the NCEP GFS and ERA5, respectively. Most clustering strategies yield similar improvements over the full ensemble mean and are similar in probabilistic skill except that 1) intensity displacement space gives lower MSLP displacement and intensity errors; and 2) Euclidean space and agglomerative hierarchical clustering, when using either full or average linkage, struggle to produce reasonably sized clusters. Applying clusters derived from MSLP to 12-h APCP forecasts is not as skillful as clustering by 12-h APCP directly, especially if several members contain little precipitation. Use of the same cluster membership for one lead time to cluster the forecast at another lead time is less skillful than clustering independently at each forecast lead time. Finally, the number of members within each cluster does not necessarily correspond with the best forecast, especially at the longer lead times, when the probability of the smallest cluster being the best scenario was usually underestimated. Significance Statement Numerical weather prediction ensembles are widely used, but more postprocessing tools are necessary to help forecasters interpret and communicate the possible outcomes. This study evaluates various clustering approaches, combining a large number of model forecasts with similar attributes together into a small number of scenarios. The 1–9-day forecasts of both sea level pressure and 12-h precipitation are used to evaluate the clustering approaches for a large number of U.S. East Coast winter cyclones, which is an important forecast problem for this region.

Probabilistic seasonal precipitation forecasts using quantiles of ensemble forecasts

Seasonal precipitation forecasting is vital for weather-sensitive sectors. Global Circulation Models (GCM) routinely produce ensemble Seasonal Climate Forecasts (SCFs) but suffer from issues like low forecast resolution and skills. To address these issues in this study, we introduce a post-processing method, Quantile Ensemble Bayesian Model Averaging (QEBMA). It utilises quantiles from a GCM ensemble forecast to create a pseudo-ensemble forecast. Through their reasonable linear relationships with observations, each pseudo-member connects a hurdle distribution with a point mass at zero for dry months and a gamma distribution for wet months. These distributions are mixed to construct a forecast probability distribution with their weights, proportional to the quantiles’ historical forecast performance. QEBMA is applied to three GCMs, including GloSea5 from the United Kingdom, ECMWF from Europe and ACCESS-S1 from Australia, for monthly precipitation forecasts in 32 locations across four climate zones in Australia. Leave-one-month-out cross-validation results illustrate that QEBMA enhances forecast skills compared to raw GCMs and other post-processing techniques, including quantile mapping and Extended Copula Post-Processing (ECPP), for forecast lead time of 0 to 2 months, based on five metrics. The skill improvements achieved by QEBMA are often statistically significant, particularly when compared to raw GCM forecasts across the 32 study locations. Among these post-processing models, only QEBMA consistently outperforms the SCF benchmark climatology, offering a promising alternative for improving seasonal precipitation forecasts.

Superior performance of hybrid model in ungauged basins for real-time hourly water level forecasting – A case study on the Lancang-Mekong mainstream

In the context of climate change and human influence, timely and reliable information about water level variations is crucial for downstream flood control, navigation, and water resource management, particularly for the Lancang-Mekong River, the largest transboundary river in Southeast Asia. However, accurate real-time water level forecasts remain challenging, especially in data-limited regions. This study focused on the Chiang Saen Station, the uppermost Mekong River hydrological station, comparing three methods to predict 48-hour water levels: the Variable Infiltration Capacity (VIC) and Hydrology-Hydraulic (HH) physics-based models, the Gated Recurrent Unit (GRU) model, and hybrid models (VIC-GRU and HH-GRU). Evaluations were conducted at the 1-h, 3-h, 6-h, 12-h, 24-h, and 48-h forecast lead times. Assessments at various lead times showed that hybrid models incorporating physics-based mechanisms significantly enhanced predictions. VIC-GRU demonstrated superior performance, with KGE values surpassing 0.94, NSE values exceeding 0.95, MAE below 0.17 m, and PBIAS within ± 1 % across all lead times. Compared to GRU, PBIAS decreased by 84 %, while MAE dropped by 76 % to 81 % relative to VIC and HH, respectively. Notably, seasonal variations affected hybrid model performance, especially in the dry season. Optimal parameter selections enhanced model accuracy by 48 % to 49 %. These results underscore the potential of combining physics-based and deep learning models for more accurate, high-temporal-resolution real-time water level forecasts in data-scarce regions. A comprehensive study of streamflow characteristics enhances the advantages of this integration. This research contributes valuable insights to water level prediction in data-scarce regions and informs flood risk reduction and management in the Lancang-Mekong mainstream.

Prediction of Tropical Cyclogenesis Based on Machine Learning Methods and Its SHAP Interpretation

AbstractThis study trains three machine learning models with varying complexity—Random Forest, Support Vector Machine, and Neural Network—to predict cyclogenesis at a forecast lead time of 24 hr for given tropical disturbances identified by an optimized Kalman Filter algorithm. The overall performance is competent in terms of f1‐scores (∼0.8) compared to previous research of the same kind. An assessment by SHapley Additive exPlanations (SHAP) values reveals that mid‐level (500 hPa) vorticity is the most influential factor in deciding if a tropical disturbance is developing or non‐developing for all three models. Wind shear and tilting are found to hold a certain level of importance as well. These results encourage further experiments that use physical models to explore the dynamical, mid‐level pathway to tropical cyclogenesis. Another usage of SHAP values in this work is to explain how a machine learning model decides if an individual tropical disturbance case will develop, by listing the contribution of each feature to the output genesis probability, illustrated by a case study of Typhoon Halong. This increases the reliability of the machine learning models, and forecasters can take advantage of such information to issue tropical cyclone formation warnings more accurately. Several caveats of the current machine learning application in the studies of tropical cyclogenesis are discussed and can be considered for future research. These can benefit the interpretation and emphasis of certain output fields in the operational dynamical prediction system, which can contribute to more timely cyclogenesis forecasts.

Technical Note: Resolution enhancement of flood inundation grids

Abstract. High-resolution flood maps are needed for more effective flood risk assessment and management. Producing these directly with hydrodynamic models is slow and computationally prohibitive at large scales. Here we demonstrate a new algorithm for post-processing low-resolution inundation layers by using high-resolution terrain models to disaggregate or downscale. The new algorithm is roughly 8 times faster than state-of-the-art algorithms and shows a slight improvement in accuracy when evaluated against observations of a recent flood using standard performance metrics. Qualitatively, the algorithm generates more physically coherent flood maps in some hydraulically challenging regions compared to the state of the art. The algorithm developed here is open source and can be applied in conjunction with a low-resolution hydrodynamic model and a high-resolution DEM to rapidly produce high-resolution inundation maps. For example, in our case study with a river reach of 20 km, the proposed algorithm generated a 4 m resolution inundation map from 32 m hydrodynamic model outputs in 33 s compared to a 4 m hydrodynamic model runtime of 34 min. This 60-fold improvement in runtime is associated with a 25 % increase in RMSE when compared against the 4 m hydrodynamic model results and observations of a recent flood. Substituting downscaling into flood risk model chains for high-resolution modelling has the potential to drastically improve the efficiency of inundation map production and increase the lead time of impact-based forecasts, helping more at-risk communities prepare for and mitigate flood damages.

Adapting subseasonal-to-seasonal (S2S) precipitation forecast at watersheds for hydrologic ensemble streamflow forecasting with a machine learning-based post-processing approach

Accurate and reliable precipitation predictions made by dynamical forecast models could provide crucial information for human socioeconomic activities by enabling hydrologic forecasts at the Subseasonal-to-Seasonal (S2S) timescale. To utilize available S2S precipitation predictions for hydrologic forecasts, post-processing techniques have been applied to adapt the raw S2S precipitation to local watersheds. However, conventional statistical-based post-processing techniques are more focused on correcting the forecast bias, but rather limited in improving the predictive skill of available S2S precipitation forecasts. In this study, we combine the Random Forest classifiers (RF) with the Bias Correction and Spatial Disaggregation (BCSD) to adapt the 10-member ensemble precipitation forecast from the NASA Goddard Earth Observing System model version 5 (GEOS5) at 4 watersheds located in the NCEI South climate region of the United States. The adapted S2S precipitation is further applied for streamflow forecast by forcing a classical lumped hydrologic model. The performance of S2S precipitation as well as the corresponding streamflow predictions are benchmarked with the randomly resampled precipitation and the corresponding Ensemble Streamflow Prediction (ESP) framework-generated streamflow predictions. Evaluation statistics of Kling-Gupta Efficiency (KGE), Continuous Ranked Probability Skill Score (CRPSS), Reliability, Resolution, and Sharpness are employed to evaluate the predictive skill of precipitation and streamflow both deterministically and probabilistically. Our results indicate that dynamical S2S precipitation after forecast adaptation leads to consistently higher deterministic skill over ESP at all forecast lead times and across study watersheds. However, at longer forecast lead times beyond 10–15 days, S2S precipitation with a limited ensemble size does not present higher probabilistic skill than ESP. Our results shows that the joint application of RF and BCSD improves the predictive skill of the raw S2S precipitation at study watersheds in contrast to BCSD. Further, the added predictive skill of S2S precipitation brought by RF propagates into streamflow predictions, predominantly at longer forecast lead times exceeding 10 days. Overall, our results highlight the potential success of future work to apply other data-driven approaches to adapt the raw precipitation to local watersheds for more accurate and reliable streamflow forecasts at the S2S timescale.

A new statistical downscaling approach for short‐term forecasting of summer air temperatures through a fusion of deep learning and spatial interpolation

AbstractReliable early forecasting of extreme summer air temperatures is essential for effectively managing and mitigating the socioeconomic damage caused by thermal disasters. Numerical weather prediction models have become valuable tools for forecasting air temperature; however, they incur high computational costs, resulting in coarse spatial resolution and systematic bias owing to imperfect parametrization. To address these problems, we developed a novel statistical downscaling and bias correction method (named DeU‐Net) for the maximum and minimum air temperature (Tmax and Tmin respectively) forecasts obtained from the Global Data Assimilation and Prediction System with a spatial resolution of 10 to 1.5 km over South Korea through the fusion of deep learning (i.e., U‐Net) and spatial interpolation. In this study, we used a methodology to decompose statistically downscaled Tmax and Tmin forecasts into temporal dynamics over South Korea and spatial fluctuations by pixels. When comparing the proposed DeU‐Net with the dynamical downscaling model (i.e., Local Data Assimilation and Prediction System) and support vector regression‐based statistical downscaling model at the seen and unseen stations for forecasting the next‐day Tmax and Tmin, DeU‐Net showed the highest spatial correlation and the lowest root‐mean‐square error in all cases. In a qualitative evaluation, DeU‐Net successfully produced a detailed spatial distribution most similar to the observations. A further comparison extending the forecast lead time to 7 days indicated that the proposed DeU‐Net is a better downscaling approach than support vector regression, regardless of the forecast lead time. These results demonstrate that bias‐corrected high spatial resolution air temperature forecasts with relatively long forecast lead times in summer can be effectively produced using the proposed model for operational forecasting.

Predicting Tropical Cyclone Formation with Deep Learning

Abstract Exploring new techniques to improve the prediction of tropical cyclone (TC) formation is essential for operational practice. Using convolutional neural networks, this study shows that deep learning can provide a promising capability for predicting TC formation from a given set of large-scale environments at certain forecast lead times. Specifically, two common deep-learning architectures including the residual net (ResNet) and UNet are used to examine TC formation in the Pacific Ocean. With a set of large-scale environments extracted from the NCEP–NCAR reanalysis during 2008–21 as input and the TC labels obtained from the best track data, we show that both ResNet and UNet reach their maximum forecast skill at the 12–18-h forecast lead time. Moreover, both architectures perform best when using a large domain covering most of the Pacific Ocean for input data, as compared to a smaller subdomain in the western Pacific. Given its ability to provide additional information about TC formation location, UNet performs generally worse than ResNet across the accuracy metrics. The deep learning approach in this study presents an alternative way to predict TC formation beyond the traditional vortex-tracking methods in the current numerical weather prediction. Significance Statement This study presents a new approach for predicting tropical cyclone (TC) formation based on deep learning (DL). Using two common DL architectures in visualization research and a set of large-scale environments in the Pacific Ocean extracted from the reanalysis data, we show that DL has an optimal capability of predicting TC formation at the 12–18-h lead time. Examining the DL performance for different domain sizes shows that the use of a large domain size for input data can help capture some far-field information needed for predicting TCG. The DL approach in this study demonstrates an alternative way to predict or detect TC formation beyond the traditional vortex-tracking methods used in the current numerical weather prediction.

A seamless blended multi‐model ensemble approach to probabilistic medium‐range weather pattern forecasts over the UK

AbstractThis paper describes a new seamless blended multi‐model ensemble configuration of an existing probabilistic medium‐ to extended‐range weather pattern forecasting tool (called Decider) run operationally at the Met Office. In its initial configuration, the tool calculated and presented probabilistic weather pattern forecast information for five individual ensemble forecasting systems, which varied in terms of their number of ensemble members, horizontal resolution, update frequencies and forecast lead time. This resulted in multiple forecasts for the same validity time which varied in terms of forecast skill depending on the lead time in question. This presented challenges for end‐users (e.g., operational meteorologists) in terms of knowing which forecast output is best to use and at which lead time, as well as knowing what to do in situations where forecasts varied between ensembles. To get around these challenges, a new seamless blended multi‐model ensemble configuration has been implemented operationally, comprising of output from five separate ensembles, and provides a single best forecast from day one out to day 45. Objective verification for a set of eight weather pattern groups covering forecasts initialized over a 6‐year period (2017–2022) shows that the seamless blended multi‐model ensemble forecasts are at least as good as, if not better than the best performing individual model.

Assessing RRFS versus HRRR in Predicting Widespread Convective Systems over the Eastern CONUS

Abstract This study provides a comparison of the operational HRRR version 4 and its eventual successor, the experimental Rapid Refresh Forecast System (RRFS) model (summer 2022 version), at predicting the evolution of convective storm characteristics during widespread convective events that occurred primarily over the eastern United States during summer 2022. In total 32 widespread convective events were selected using observations from the MRMS composite reflectivity, which includes an equal number of MCSs, quasi-linear convective systems (QLCSs), clusters, and cellular convection. Each storm system was assessed on four primary characteristics: total storm area, total storm count, storm area ratio (an indicator of mean storm size), and storm size distributions. It was found that the HRRR predictions of total storm area were comparable to MRMS, while the RRFS overpredicted total storm area by 40%–60% depending on forecast lead time. Both models tended to underpredict storm counts particularly during the storm initiation and growth period. This bias in storm counts originates early in the model runs (forecast hour 1) and propagates through the simulation in both models indicating that both miss storm initiation events and/or merge individual storm objects too quickly. Thus, both models end up with mean storm sizes that are much larger than observed (RRFS more so than HRRR). Additional analyses revealed that the storm area and individual storm biases were largest for the clusters and cellular convective modes. These results can serve as a benchmark for assessing future versions of RRFS and will aid model users in interpreting forecast guidance.

To Wait or Not to Wait: Optimal Ordering Under Lead Time Uncertainty and Forecast Updating

To Wait or Not to Wait: Optimal Ordering Under Lead Time Uncertainty and Forecast Updating

A Machine Learning Augmented Data Assimilation Method for High‐Resolution Observations

AbstractThe accuracy of initial conditions is an important driver of the forecast skill of numerical weather prediction models. Increases in the quantity of available measurements, particularly high‐resolution remote sensing observational data products from satellites, are valuable inputs for improving those initial condition estimates. However, the traditional data assimilation methods for integrating observations into forecast models are computationally expensive. This makes incorporating dense observations into operational forecast systems challenging, and it is often prohibitively time‐consuming. Additionally, high‐resolution observations often have correlated observation errors which are difficult to estimate and create problems for assimilation systems. As a result, large quantities of data are discarded and not used for state initialization. Using the Lorenz‐96 system for testing, we demonstrate that a simple machine learning method can be trained to assimilate high‐resolution data. Using it to do so improves both initial conditions and forecast accuracy. Compared to using the Ensemble Kalman Filter with high‐resolution observations ignored, our augmented method has an average root‐mean‐squared error reduced by 37%. Ensemble forecasts using initial conditions generated by the augmented method are more accurate and reliable at up to 10 days of forecast lead time.

Study and application on the optimal quantile forecast of precipitation in an ensemble forecast system

AbstractQuantiles of precipitation are widely used in ensemble forecast systems. At present, the common practice is to provide precipitation amounts corresponding to different quantiles to users directly, which will make it difficult for users to extract reliable forecast information. Therefore, this study investigates the statistically optimal (using threat score (TS) as a metric) quantiles of precipitation in an ensemble forecast system constructed using the WRF V4.0 model. The main conclusions are as follows: The threat‐score‐optimal quantiles for light rain, moderate rain, heavy rain, rainstorm, and heavy rainstorm forecasts are 40%–60%, 60%–70%, 60%–80%, 70%–80%, and 80%, respectively. Overall, the optimal quantile increases with the rise in precipitation magnitude or the extension of forecast lead time. All the optimal quantile forecast products have higher TS than the corresponding control forecast, ensemble mean forecast, and probability‐matched ensemble mean forecast products. The merged threat‐score‐optimal quantile forecast product formed by combining the optimal quantile forecasts of different precipitation magnitudes shows obvious advantages over other products in statistical verification and case studies, and it shows good potential to be operationally implemented in the future.